Classification of eye diseases in fundus images using Convolutional Neural Network (CNN) method with EfficientNet architecture

Abstract

This research is designed to classify eye disease conditions into three classes, namely normal, cataract, and glaucoma using a system. The system will use Convolutional Neural Network (CNN) with EfficientNet architecture. The EfficientNet that will be used is EfficientNet-B0. The dataset in this paper is obtained from Kaggle totaling 300 images. From this data, augmentation is carried out so that 3.600 images are obtained consisting of "normal" (1200 images), "cataract" (1200 images), and "glaucoma" (1200 images). This data is processed into 4 different datasets, namely the original dataset, augmentation dataset, augmentation dataset that has been preprocessed grayscale, and augmentation dataset that has been preprocessed thresholding. The best results are obtained using the Adam optimizer, learning rate 0.00001, and batch size 32, and iterations of 20 epochs. The best dataset is the augmentation dataset that has been preprocessed grayscale with an accuracy of 79.22%, precision value of 80.3%, recall value of 79.22%, F1-Score of 78.87%.